1 Reply Latest reply on Oct 11, 2012 11:29 AM by dougboutwell

    Apples to apples - how can I make sense of DNGPE's color readouts?

    dougboutwell

      The simple version of this question is "how do I convert a DNG file such that Photoshop's HSV (and CIELAB) readouts agree with DNGPE's?"  Put another way, I want to have the same DNG image open in Photoshop and DNGPE, and have both applications agree on the color numbers for a particular chart patch.

       

      It seems like using ACR to convert an image into ProPhotoRGB would get me close, at least in the hue and sat dimensions, but that isn't the case in practice. 

      Worse yet, the L*a*b* numbers disagree, and as far as I understand, they shouldn't.  The closest I've been able to get is using dcraw on the command line to spit out linear 16-bit TIFFs in ProPhotoRGB, but it's still not exact, and it strikes me as weird that I'd get closer with an open-source tool than Camera Raw.

       

      I'd like to just derive my list of HSV vectors from Photoshop and plug that list into a DNGPE recipe (rather than hand-tweaking color patches using the DNGPE GUI), but if I can't get Photoshop and DNGPE to agree on the source color, this becomes impossible, as I'd be comparing apples to oranges.

       

      Also, it's my understanding that the DNG Profile HueSatMaps are based on HSV, but DNGPE seems to specify its list of corrections as HSL... is that correct?  Any particular reason that the corrections are a list of HSL vectors that are later translated into HSV for the profile?

       

      Thanks in advance!

        • 1. Re: Apples to apples - how can I make sense of DNGPE's color readouts?
          dougboutwell Level 1

          Following up to myself here - I was able to get Camera Raw to convert an image such that the L*a*b* numbers matched DNGPE, but Hue and Sat are still off.  What I WAS doing previously, to no avail, was converting in Camera Raw using the same BASE profile I had selected in DNGPE.  I thought that since I hadn't made any corrections or changes in DNGPE, it would only be applying the base profile, and the images should match.

           

          To fix that, I used DNGPE to export a new profile (again, with no actual edits made), and converted using the exported profile.  Now Photoshop and DNGPE agree about the L*a*b* figures at least, though they still disagree about hue / sat.  DNGPE consistently reports hue numbers about 6º-10º higher than Photoshop, and saturation numbers that are 50-100% higher.  Saturation difference, I suspect, are from comparing HSB in Photoshop to HSL in DNGPE.  Not sure what would be causing the hue inaccuracy.  The RGB numbers differ wildly, too, so this is probably a color management issue.

           

          Duplicating DNGPE's color management to obtain identical Hue and Sat figures in another application is now the problem.  Put as simply as possible - "Where does DNGPE get it's hue / sat figures?"

           

          Message was edited by: dougboutwell